An unmanned aerial vehicle online real-time lightweight electromagnetic sensitive regulation and control device

Abstract

The present application relates to a kind of unmanned aerial vehicle online real-time lightweight electromagnetic sensitive regulation and control device, device includes data acquisition analysis module: airborne development board real-time acquisition unmanned aerial vehicle flight data in flight process, the flight data obtained is input to PS-BiLSTM regulation and control model, PS-BiLSTM regulation and control model real-time analysis whether flight data appears electromagnetic sensitive, and output normal state estimate value;Online real-time regulation and control module: according to normal state estimate value, the electromagnetic interference influenced electromagnetic sensitive flight data of current time is regulated and controlled.The present application can be for electromagnetic interference to cause sensitive flight data to carry out online real-time lightweight regulation and control and restore to normal data, reduce the influence of electromagnetic sensitive on unmanned aerial vehicle flight state perception and control process, improve the task efficiency of unmanned aerial vehicle in actual electromagnetic environment.

Description

Technical Field

[0001] This invention relates to the field of electromagnetic detection, and in particular to an online, real-time, lightweight electromagnetic sensing control device for unmanned aerial vehicles (UAVs). Background Technology

[0002] When drones fly in real-world electromagnetic environments, their onboard electronic equipment is susceptible to electromagnetic interference, causing sensitive fluctuations in output flight data. This further affects the flight control system's accurate perception of the drone's status, thereby reducing the effectiveness of drone flight missions. Therefore, online real-time control of flight data affected by electromagnetic interference is crucial for ensuring the safe flight of drones in real-world electromagnetic environments.

[0003] Currently, existing technologies for addressing electromagnetic interference (EMI) issues in drones primarily employ electromagnetic shielding, filtering, and grounding to reduce its impact. However, these methods mainly optimize the propagation path of EMI or the hardware's anti-interference capabilities. When EMI has already affected airborne electronic equipment and caused sensitive fluctuations in output data, it is difficult to implement real-time electromagnetic control measures. Furthermore, drones in actual flight are characterized by limited computing resources and high response time requirements. Therefore, to address the EMI-induced electromagnetic sensitivity fluctuations in drone flight data, there is an urgent need for an electromagnetic sensitivity control device that can be deployed on the drone's airborne platform, possesses real-time computing capabilities, and has low computational overhead. This device would enable rapid recovery and dynamic correction of EMI-sensitive data during flight, thereby improving the drone's continuous flight capability and autonomous support level in actual electromagnetic environments. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art and provide an online real-time lightweight electromagnetic sensitivity control device for unmanned aerial vehicles, thus solving the deficiencies of the prior art.

[0005] The objective of this invention is achieved through the following technical solution: an online real-time lightweight electromagnetic sensitivity control device for unmanned aerial vehicles (UAVs), the device comprising a control model deployment and configuration module, a data acquisition and analysis module, an online real-time control module, and a judgment module;

[0006] The control model deployment and configuration module: mounts the trained PS-BiLSTM control model on the airborne development board and integrates it into the UAV platform, establishing a data transmission link between the data acquisition and analysis module and the flight control system;

[0007] The data acquisition and analysis module: The airborne development board acquires flight data during the flight of the UAV in real time, inputs the acquired flight data into the PS-BiLSTM control model, and the PS-BiLSTM control model analyzes in real time whether electromagnetic susceptibility occurs in the flight data and outputs the estimated value of the normal state.

[0008] The online real-time control module: adjusts the electromagnetically sensitive flight data affected by electromagnetic interference at the current moment according to the normal state estimate, generates corresponding recovery data, and sends the adjusted flight data to the UAV flight data management system. The UAV platform completes the attitude, position and motion state control of the UAV based on the corrected flight data.

[0009] The judgment module determines whether the UAV platform has completed its flight mission. If not, the data acquisition and analysis module continues to collect new flight status data and repeats the data prediction, recovery, and feedback process to achieve continuous online real-time electromagnetic sensitivity control of the UAV in the actual electromagnetic environment.

[0010] The PS-BiLSTM regulation model includes a stacked BiLSTM and a principal component analysis unit.

[0011] The stacked BiLSTM consists of two layers of BiLSTM network. The first layer of BiLSTM is used to extract the basic temporal features of the input data, and the second layer of BiLSTM fuses the high-level temporal features to realize the modeling of the dynamic change law of flight data.

[0012] The principal component analysis unit reduces the dimensionality of the weight matrix in the BiLSTM network. By retaining the main feature information in the weight matrix, it reduces the model parameter size and computational complexity, thereby reducing the model's computational resource consumption while maintaining the model's ability to extract temporal features of flight data.

[0013] The device also includes an evaluation module, which uses mean square error, mean absolute error and coefficient of determination as evaluation indicators to evaluate the electromagnetic sensitivity control of flight parameter data.

[0014] Mean square error and mean absolute error are used to evaluate the applicability of the output flight data after electromagnetic sensitivity control to the real data, while the coefficient of determination is used to characterize the degree of fit between the flight data after electromagnetic sensitivity control and the real data.

[0015] The device also includes a key flight data determination module, which selects a target UAV platform, analyzes the key airborne electronic devices and their output data types involved in the platform's flight, and determines the key flight data that affects the UAV's state perception and control after being affected by electromagnetic interference, as the data object for subsequent electromagnetic sensitivity control.

[0016] The device also includes a flight test module:

[0017] The flight test module is responsible for: conducting UAV simulation flight tests to acquire operational data of the UAV under typical flight trajectories and different flight states; conducting normal flight tests of the UAV and flight tests under electromagnetic interference environments; collecting flight data of the UAV under stable operating conditions through normal flight tests to establish normal variation patterns of flight data; and collecting flight data of abnormal changes caused by interference to airborne electronic equipment by applying external electromagnetic interference signals.

[0018] The device also includes an electromagnetic sensitivity adjustment dataset construction module;

[0019] The electromagnetic sensitivity adjustment dataset construction module preprocesses the acquired UAV flight data and, based on the correspondence between normal flight data and disturbed flight data, constructs a dataset containing input electromagnetic sensitivity data and target control data for training the PS-BiLSTM control model.

[0020] The device also includes a model training module;

[0021] The model training module inputs the constructed electromagnetic sensitivity control dataset into the PS-BiLSTM control model for training. By iteratively optimizing the network parameters, the model learns the temporal mapping relationship between electromagnetically disturbed flight data and normal flight data. When the model reaches the preset training accuracy, it outputs a PS-BiLSTM control model that meets the online real-time control requirements of the UAV.

[0022] The present invention has the following advantages: an online real-time lightweight electromagnetic sensitivity control device for unmanned aerial vehicles (UAVs) can perform online real-time lightweight control of sensitive flight data caused by electromagnetic interference and restore it to normal data, thereby reducing the impact of electromagnetic sensitivity on the UAV's flight status perception and control process and improving the UAV's mission performance in actual electromagnetic environments. Attached Figure Description

[0023] Figure 1 This is a schematic diagram illustrating the process of implementing the device of the present invention.

[0024] Figure 2 This is a schematic diagram of an electromagnetic susceptibility that jams during roll angle.

[0025] Figure 3 This is a diagram showing the results of the present invention's sensitive control of roll angle data affected by electromagnetic interference. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the detailed description of the embodiments of this application provided below with reference to the accompanying drawings is not intended to limit the scope of protection of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application. The present invention will be further described below with reference to the accompanying drawings.

[0027] This invention specifically relates to an online real-time lightweight electromagnetic sensitivity control device for unmanned aerial vehicles (UAVs). Addressing the sensitive flight data generated by UAV onboard electronic equipment after being affected by electromagnetic interference, this device can effectively control the electromagnetically sensitive fluctuations in UAV flight data in real time. Simultaneously, this method effectively reduces the demand for computing resources, meeting the online deployment requirements of UAV onboard platforms.

[0028] like Figure 1 As shown, this invention includes a key flight data determination module, a flight test module, an electromagnetic sensitivity adjustment dataset construction module, a model training module, a control model deployment and configuration module, a data acquisition and analysis module, an online real-time control module, an evaluation module, and a judgment module. The specific implementation process is as follows:

[0029] Step 1: Key Flight Data Determination Module. This module determines the UAV platform and key flight data. First, the target UAV platform is selected, and the key airborne electronic devices involved in the platform's flight and their output data types are analyzed. The key flight data that will affect the UAV's state perception and control after being affected by electromagnetic interference is determined, and these data will be used as the data objects for subsequent electromagnetic sensitivity control.

[0030] Step Two: The flight test module acquires flight data of the UAV under normal conditions and under electromagnetic interference. First, simulated flight tests of the UAV are conducted to acquire operational data of the UAV on typical flight trajectories and under different flight conditions. Further, normal flight tests and flight tests under electromagnetic interference environments are carried out. Specifically, flight data under stable operating conditions is collected through normal flight tests to establish the normal variation patterns of flight data; by applying external electromagnetic interference signals, flight data showing abnormal changes in airborne electronic equipment after interference is collected to construct flight data affected by electromagnetic interference.

[0031] Furthermore, in constructing flight data affected by electromagnetic interference, this invention uses a stuck-type electromagnetic susceptibility as a typical form of electromagnetic susceptibility, while stuck-type electromagnetic susceptibility represents normal flight parameter data. exist It becomes a fixed value after a certain time. This results in stuck electromagnetic sensitive data. . For data sensitivity description of stuck conditions, among which It is a small random noise signal.

[0032] like Figure 2 The diagram shows an electromagnetic sensitivity pattern where the roll angle becomes stuck. When affected by the electromagnetic environment, the roll angle output data becomes stuck during electromagnetic interference and cannot be updated normally as the UAV's flight status changes.

[0033] Step 3: The electromagnetic sensitivity adjustment dataset construction module preprocesses the acquired UAV flight data, including data cleaning, standardization, and time series reconstruction. Based on the correspondence between normal flight data and disturbed flight data, a training sample containing input electromagnetic sensitivity data and target control data is constructed, enabling the model to learn the changing patterns of flight data under electromagnetic interference and the mapping relationship to the normal state.

[0034] Furthermore, the flight data is first standardized. The original flight dataset is represented as follows: ,in Indicates input variables, Indicates the output variable. Represents the real number threshold. It is the dimension of the input features. This represents the total number of samples. The Z-Score normalization method was applied to the original flight dataset, and the calculation is as follows:

[0035] ,

[0036] in, yes The average vector of flight data, and for The standard deviation of the flight data. Substitute into the formula Generate a standardized flight dataset Based on this, a phase space reconstruction method is used to reconstruct the dataset. A sliding window reconstruction is performed to construct the input and output datasets suitable for a single-step predictive regression model. The phase space reconstruction is shown below:

[0037] ,

[0038] Given ,in This represents standardized flight data. Represents the first in the dataset One sample, dimension Corresponding to the time step parameter in the LSTM model, It's a time delay. In the case of single-step prediction... At the same time, each The corresponding prediction target is constructed as follows:

[0039] ,

[0040] in, Indicates the first Standardized target value for each sample. Delay time. Set to 1, and dimension The determination should be based on the specific application. A regression modeling dataset is obtained by reconstructing standardized flight data.

[0041] .

[0042] Step four, the model training module, will train the PS-BiLSTM control model. The completed electromagnetic sensitivity control dataset will be input into the PS-BiLSTM network model for training. By iteratively optimizing the network parameters, the model will learn the temporal mapping relationship between flight data subject to electromagnetic interference and normal flight data. Once the model reaches the preset training accuracy, it will output a PS-BiLSTM electromagnetic sensitivity control model that meets the requirements for online real-time control of UAVs.

[0043] The PS-BiLSTM model first employs a bidirectional long short-term memory network structure (BiLSTM) to extract the temporal correlation features of flight data by simultaneously utilizing both forward and backward temporal information.

[0044] Furthermore, this invention proposes a stacked BiLSTM, which consists of two BiLSTM layers. The first BiLSTM layer is used to extract the basic temporal features of the input data, while the second BiLSTM layer further fuses higher-level temporal features to model the dynamic changes in flight data. Since the structure of each LSTM layer is the same, the following explanation uses the feedforward LSTM network in the first BiLSTM layer as an example to illustrate its computation process.

[0045] Assuming at time step The input to the regression model at that location is The corresponding target output is The number of units in an LSTM layer is First, the output of each gate is calculated as follows:

[0046] ,

[0047] ,

[0048] ,

[0049] ,

[0050] Update cell status:

[0051] ,

[0052] in, , , These are the forget gate output, weights, and biases. It is an activation function. It is a time step The hidden state; , , These are the input gate output, weights, and biases, respectively. , , These are the output gate output, weights, and biases; It is input information. , These are the input information weights and biases; This represents the Hadamard product. Finally, by using the formula... and formula Substitute into the following formula to calculate the output value at the current time step:

[0053] ,

[0054] Compared to traditional unidirectional LSTM, BiLSTM, in detecting electromagnetic susceptibility in UAVs, can more comprehensively capture the temporal correlation characteristics of anomalous signals by simultaneously learning the forward and backward temporal features of time-series data. Firstly, in terms of feature extraction capabilities, the bidirectional structure of BiLSTM can simultaneously analyze historical states and future trends, effectively overcoming the limitation of unidirectional LSTM, which can only acquire past information. This function makes it particularly suitable for detecting electromagnetic susceptibility in data with delay effects, such as situations where the initial features of electromagnetic susceptibility are weak but the subsequent impact is significant. Furthermore, in practical applications, BiLSTM can effectively reduce the false alarm rate of the model through complementary verification of bidirectional information, which is crucial for UAV flight safety.

[0055] Furthermore, to meet the resource constraints of online deployment under UAV airborne conditions, the stacked BiLSTM network is lightweighted. This invention employs Principal Component Analysis (PCA) to reduce the dimensionality of the weight matrix in the BiLSTM network. By retaining the main feature information in the weight matrix, the model parameter size and computational complexity are reduced. While reducing the network's computational resource consumption, the model's ability to extract temporal features of flight data is maintained, ultimately forming a PS-BiLSTM electromagnetic sensitivity control model. Taking the weight matrix of the BiLSTM layer as an example, the reorganized weight matrix is ​​centered, and the results are as follows:

[0056] ,

[0057] in, express The mean vector of each row in the matrix. Calculate the covariance matrix. for:

[0058] ,

[0059] Singular value decomposition is introduced to obtain the covariance matrix. The eigenvalues ​​and eigenvectors of the matrix. Covariance matrix. The decomposition is as follows:

[0060] ,

[0061] in, Covariance matrix The eigenvalues. By analyzing the eigenvalues ​​of the decomposition... Sort in descending order to remove some eigenvectors with relatively small weights, thus reducing the dimensionality of the weight matrix. The eigenvalue set sorted in descending order is represented as:

[0062] ,

[0063] The eigenvalues ​​after dimensionality reduction and the eigenvalues ​​before dimensionality reduction must satisfy the following:

[0064] ,

[0065] in, This is the threshold for structured pruning of the model, typically ranging from [value missing]. .

[0066] The electromagnetic sensitivity control of flight parameter data was evaluated using an evaluation module, employing mean square error (MSE), mean absolute error (MAE), and coefficient of determination. As evaluation metrics, MSE, MAE, and The calculation formula is as follows:

[0067] ,

[0068] ,

[0069] ,

[0070] in, This represents actual flight parameter data. This indicates flight data after electromagnetic control. This represents the average value of the actual flight parameter data. MSE and MAE are used to evaluate the applicability of the proposed method's output flight data after electromagnetic sensitivity control to real data. It is used to characterize the degree of fit between flight data after electromagnetic sensitivity control and real data.

[0071] Step 5: The control model deployment and configuration module integrates the airborne development board carrying the PS-BiLSTM control model into the UAV platform and establishes a data transmission link between it and the data acquisition and analysis module and the flight control system. The control device is used to acquire UAV operating status data in real time and generate corresponding data correction information based on the model output results.

[0072] Step Six: The airborne development board of the data acquisition and analysis module acquires flight data of the UAV in real time during flight and inputs the acquired flight data into the PS-BiLSTM control model. The PS-BiLSTM control model analyzes the flight data in real time to determine if electromagnetic susceptibility occurs and outputs an estimate of the normal state.

[0073] Step 7: The online real-time control module adjusts the electromagnetically sensitive flight data affected by electromagnetic interference at the current moment based on the normal state estimate output by the electromagnetically sensitive control device, and generates corresponding recovery data.

[0074] The recovered data characterizes the normal operation of the UAV under conditions where electromagnetic sensitivity does not occur. The adjusted flight data is then sent to the UAV flight data management system. Based on the corrected flight status information, the UAV platform can control the UAV's attitude, position, and motion, reducing the impact of electromagnetic sensitivity of onboard electronic equipment on flight safety.

[0075] Step 8: The judgment module determines whether the UAV platform has completed its flight mission. If not, it returns to Step 6 to continue collecting new flight status data and repeats the data prediction, recovery, and feedback process to achieve continuous online real-time electromagnetic sensitivity control of the UAV in the actual electromagnetic environment. If yes, the operation ends.

[0076] like Figure 3As shown in the figure, the electromagnetically sensitive flight data after being subjected to electromagnetic interference can be effectively restored to its normal trend after being controlled by the PS-BiLSTM model, realizing online real-time control of electromagnetically sensitive flight data. Experimental results show that the electromagnetic sensitivity control performance of this invention is MSE=0.0231, MAE=0.0483, and... =0.9672, and the control time required is 0.964 milliseconds. The accuracy of the proposed method is quantitatively demonstrated through the electromagnetic sensitivity control evaluation index, and the control time and the airborne equipment data update time are on the same order of magnitude, meeting the requirements for real-time electromagnetic sensitivity control.

[0077] The above description is merely a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and improvements, and can be altered within the scope of the concept described herein through the above teachings or related technologies or knowledge. Modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention should be within the protection scope of the appended claims.

Claims

1. A lightweight, online, real-time electromagnetic sensing control device for unmanned aerial vehicles (UAVs), characterized in that: The device includes a control model deployment and configuration module, a data acquisition and analysis module, an online real-time control module, and a judgment module; The control model deployment and configuration module: mounts the trained PS-BiLSTM control model on the airborne development board and integrates it into the UAV platform, establishing a data transmission link between the data acquisition and analysis module and the flight control system; The data acquisition and analysis module: The airborne development board acquires flight data during the flight of the UAV in real time, inputs the acquired flight data into the PS-BiLSTM control model, and the PS-BiLSTM control model analyzes in real time whether electromagnetic susceptibility occurs in the flight data and outputs the estimated value of the normal state. The online real-time control module: adjusts the electromagnetically sensitive flight data affected by electromagnetic interference at the current moment according to the normal state estimate, generates corresponding recovery data, and sends the adjusted flight data to the UAV flight data management system. The UAV platform completes the attitude, position and motion state control of the UAV based on the corrected flight data. The judgment module determines whether the UAV platform has completed its flight mission. If not, the data acquisition and analysis module continues to collect new flight status data and repeats the data prediction, recovery, and feedback process to achieve continuous online real-time electromagnetic sensitivity control of the UAV in the actual electromagnetic environment.

2. The online real-time lightweight electromagnetic sensing control device for unmanned aerial vehicles according to claim 1, characterized in that: The PS-BiLSTM regulation model includes a stacked BiLSTM and a principal component analysis unit. The stacked BiLSTM consists of two layers of BiLSTM network. The first layer of BiLSTM is used to extract the basic temporal features of the input data, and the second layer of BiLSTM fuses the high-level temporal features to realize the modeling of the dynamic change law of flight data. The principal component analysis unit reduces the dimensionality of the weight matrix in the BiLSTM network. By retaining the main feature information in the weight matrix, it reduces the model parameter size and computational complexity, thereby reducing the model's computational resource consumption while maintaining the model's ability to extract temporal features of flight data.

3. The online real-time lightweight electromagnetic sensing control device for unmanned aerial vehicles according to claim 1, characterized in that: The device also includes an evaluation module, which uses mean square error, mean absolute error and coefficient of determination as evaluation indicators to evaluate the electromagnetic sensitivity control of flight parameter data. Mean square error and mean absolute error are used to evaluate the applicability of the output flight data after electromagnetic sensitivity control to the real data, while the coefficient of determination is used to characterize the degree of fit between the flight data after electromagnetic sensitivity control and the real data.

4. A lightweight, online, real-time electromagnetic sensing control device for unmanned aerial vehicles (UAVs) according to any one of claims 1-3, characterized in that: The device also includes a key flight data determination module, which selects a target UAV platform, analyzes the key airborne electronic devices and their output data types involved in the platform's flight, and determines the key flight data that affects the UAV's state perception and control after being affected by electromagnetic interference, as the data object for subsequent electromagnetic sensitivity control.

5. The online real-time lightweight electromagnetic sensing control device for unmanned aerial vehicles according to claim 4, characterized in that: The device also includes a flight test module: The flight test module is responsible for: conducting UAV simulation flight tests to acquire operational data of the UAV under typical flight trajectories and different flight states; conducting normal flight tests of the UAV and flight tests under electromagnetic interference environments; collecting flight data of the UAV under stable operating conditions through normal flight tests to establish normal variation patterns of flight data; and collecting flight data of abnormal changes caused by interference to airborne electronic equipment by applying external electromagnetic interference signals.

6. The online real-time lightweight electromagnetic sensing control device for unmanned aerial vehicles according to claim 5, characterized in that: The device also includes an electromagnetic sensitivity adjustment dataset construction module; The electromagnetic sensitivity adjustment dataset construction module preprocesses the acquired UAV flight data and, based on the correspondence between normal flight data and disturbed flight data, constructs a dataset containing input electromagnetic sensitivity data and target control data for training the PS-BiLSTM control model.

7. The online real-time lightweight electromagnetic sensing control device for unmanned aerial vehicles according to claim 6, characterized in that: The device also includes a model training module; The model training module inputs the constructed electromagnetic sensitivity control dataset into the PS-BiLSTM control model for training. By iteratively optimizing the network parameters, the model learns the temporal mapping relationship between electromagnetically disturbed flight data and normal flight data. When the model reaches the preset training accuracy, it outputs a PS-BiLSTM control model that meets the online real-time control requirements of the UAV.

Images